1 Introduction

In this assignment you need to write x86-64 assembly by hand. This is not something most programmers do often, but when it comes up in the workplace being one of the few there who can do it will help your company and make you shine.

Write your code in a file named matlib.s. We provide a base file that includes the input and output interaction (in main and printNum) to test the code you will write: matlib.s. Write your code where the comments say TO DO: write this function.

2 Logistics

2.1 Writing your code

You can edit your assembly however you wish, but we strongly recommend a CLI editor.

2.2 Testing your code

We recommend using git to version and test your code.

1. Use a git project for this HW and check it out on both your laptop and the server. You can use the same project you made in Lab 01 or a new one for just this HW
2. Download the starter code into that project directory on your laptop and git add matlib.s to tell git to track it
3. Edit locally, and when you want to test it

   1. commit your edits (e.g. git commit -a -m 'I think I got product working now')

   2. push your committed edits (e.g. git push)

   3. tell the server to pull, compile, and run with ssh

      You can do this in several steps:

      1. ssh mst3k@portal.cs.virginia.edu
      2. module load clang-llvm
      3. cd cso1-code
      4. git pull
      5. clang matlib.s && ./a.out
      6. exit

      or using a one-liner to have SSH do it all for you. Note the extra source command to tell SSH to know about modules:

      ssh mst3k@portal.cs.virginia.edu 'source /etc/profile.d/modules.sh; module load clang-llvm; cd cso1-code; git pul; clang matlbi.s && a.out'

   4. You can also put all these commands in a file (pa06test.sh on MacOS, pa06test.ps1 on Windows) and run them at once (using bash pa06test.sh on MacOS, pwsh pa06test.ps1 on Windows)

      git commit -a -m 'auto-commit before testing'
      git push
      ssh mst3k@portal.cs.virginia.edu 'source /etc/profile.d/modules.sh; module load llvm-clang; cd cso1-code; git pull; clang matlib.s && ./a.out'

This way you’ll both (a) have a backup of every version you test, (b) have a copy on the server in case something happens to your laptop, and (c) know you’re using the same compiler, processor variant, etc, that we are.

If you’d rather run locally, you may, but we still encourage frequent git commit and git push to keep backups of your work.

3 Write two functions

In AT&T syntax x86-64 assembly, write two routines: one that computes the product of two numbers, and one that computes the power of two numbers.

It should be possible to assemble your code by typing clang matlib.s to generate a file, a.out, that can be run by by typing ./a.out. When run, it should¹ ask for two integers and display their product and exponent.

3.1 Product

The first subroutine, product, should compute and return the product of the two integer arguments. It must not use multiplication or division instructions. It must compute this iteratively, not recursively.

function product(x, y)
    z = 0
    while y > 0 repeat
        z += x
        y -= 1
    end while
    return z
end function

You may assume that both of the parameters are positive integers (i.e., neither negative or zero) and that the result can fit in a single 64-bit register.

3.2 Power

The second subroutine, power, should compute and return its first argument raised to the power of its second argument. It must use the product routine you wrote to do this, not x86-64 multiplication instructions or other routines you did not write. It must compute this recursively, not iteratively.

function power(x, y)
    if y == 1 then
        return x
    else
        return product(x, power(x, y-1))
    end if
end function

You may assume that both of the parameters are positive integers (i.e., neither negative or zero) and that the result can fit in a single 64-bit register.

4 Submit

Submit your assembly as matlib.s.